postgresURL
the url of the postgress server to query
postgresUser
the username to access the postgres db
postgresPassword
the postgres password
sourcePath
the source path
outputPath
the graph output path
Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/publication
${nameNode}/${outputPath}/publication
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/dataset
${nameNode}/${outputPath}/dataset
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/otherresearchproduct
${nameNode}/${outputPath}/otherresearchproduct
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/software
${nameNode}/${outputPath}/software
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/organization
${nameNode}/${outputPath}/organization
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/project
${nameNode}/${outputPath}/project
${jobTracker}
${nameNode}
${nameNode}/${sourcePath}/datasource
${nameNode}/${outputPath}/datasource
${jobTracker}
${nameNode}
eu.dnetlib.dhp.blacklist.ReadBlacklistFromDB
--hdfsPath${workingDir}/blacklist
--hdfsNameNode${nameNode}
--postgresUrl${postgresURL}
--postgresUser${postgresUser}
--postgresPassword${postgresPassword}
yarn
cluster
PrepareMergedRelation
eu.dnetlib.dhp.blacklist.PrepareMergedRelationJob
dhp-blacklist-${projectVersion}.jar
--executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--sourcePath${sourcePath}/relation
--outputPath${workingDir}/mergesRelation
--hive_metastore_uris${hive_metastore_uris}
yarn
cluster
ApplyBlacklist
eu.dnetlib.dhp.blacklist.SparkRemoveBlacklistedRelationJob
dhp-blacklist-${projectVersion}.jar
--executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--sourcePath${sourcePath}/relation
--outputPath${outputPath}/relation
--hdfsPath${workingDir}/blacklist
--mergesPath${workingDir}/mergesRelation